A State-of-the-Art Survey of Transfer Learning in Structural Health Monitoring

Deep learning technology has been applied in structural health monitoring (SHM), but in the existing structural disease detection problems, there is a lack of available data sets to effectively train the deep neural network model. Transfer learning technology can effectively alleviate the problem of lack of data sets mentioned above by introducing relevant domain knowledge into the target domain. In order to introduce transfer learning technology into SHM, this paper first introduces the related concepts and elements of transfer learning, then summarizes the application status of transfer learning combined with deep learning technology in pavement disease detection, UAVs-based automatic disease detection, concrete disease detection, bridge apparent disease detection and other fields. Finally, the prospect of transfer learning in SHM is prospected and discussed.